Founding Research Engineer

Auxo Talent; Mountain View (CA); On-site; USD 180,000 - 260,000; Full time

Benefits offered by this job

Equity

Direct line to CEO

Ownership in product

Job summary

Auxo Talent is partnering with a seed-funded Physical AI startup in Mountain View to hire their second engineer for a founding role. You’ll report directly to the CEO and CTO, owning the robot stack across multiple arms and collaborating with leadership to bring innovative capabilities to life.

You will fine-tune models on real-world data, convert live interventions into better policies, and build robust evaluations for factory deployments.

Qualifications

• Experience fine-tuning models (VLAs, diffusion policies, or world models).
• Strong Python and PyTorch skills; comfortable with ROS 2 and real hardware.
• Experience deploying robots in real-world settings is essential.
• MS or PhD with publications at CoRL, RSS, ICRA, NeurIPS or similar is a strong plus.

Responsibilities

• Own the robot stack across control, perception, teleop, and hands-off between policy and human operator.
• Fine-tune robot foundation models on real-world data collected from deployments.
• Turn interventions from live deployments into improved policies.
• Build evals that perform reliably in real factories, not only in the lab.

Skills

Fine-tuning VLAs

Python

PyTorch

ROS 2

Real hardware

Robotics deployment

Education

MS or PhD with publications

Tools

PyTorch

ROS 2

Job description

We’re partnering with a seed-funded Physical AI startup in Mountain View. They’re hiring their second engineer.

They’ve built a remote teleop platform that runs alongside the robot’s autonomy. When a robot is about to fail, a human steps in, and that correction becomes training data. The product is close to shipping, and deployments at real industrial sites are starting now.

This is a founding role. You’ll report directly to the CEO and CTO and work closely with the leadership team. They want someone who brings something new to the room, not just another pair of hands.

What you’ll do

• Own the robot stack across several industrial arms: control, perception, teleop, and the handoff between the policy and the human operator.
• Fine-tune robot foundation models (π-series, GR00T, OpenVLA) on real-world data.
• Turn interventions from live deployments into better policies.
• Build evals that hold up in real factories, not just the lab.

What they’re looking for

• Hands-on experience fine-tuning VLAs, diffusion policies or world models.
• Strong Python and PyTorch, and comfortable with ROS 2 and real hardware.
• Experience deploying robots in the real world. This matters most.
• An MS or PhD with publications at CoRL, RSS, ICRA, NeurIPS or similar is a strong plus.
• Bonus: teleop experience, or time at a top robotics or AI lab.

Why this role

• Your models run on real robots in real factories.
• Real-world data from day one.
• Real ownership, meaningful equity, and a direct line to the CEO.